My Results Vary a Lot Between Tests — Is That Normal?
Some variation is expected, especially on Wi-Fi. Here's how much is normal, what counts as concerning, and what it means if variation is large or time-dependent.
Expected variation by connection type
| Connection type | Normal variation (back-to-back) | Investigate if above |
|---|---|---|
| Wired Ethernet (same room as modem) | 2–8% | 20% |
| Wired Ethernet (through router) | 5–10% | 25% |
| Wi-Fi 5 GHz (same room) | 10–20% | 40% |
| Wi-Fi 2.4 GHz or through walls | 15–30% | 50% |
| Mobile (4G/5G) | 20–50% | 80% |
Why Wi-Fi results vary more than Ethernet
Wi-Fi is a shared medium — all devices in range compete for the same radio spectrum. Between tests, other devices on your network (or your neighbours' networks) may have started or stopped transmitting. Channel congestion, microwave interference, and even someone walking between the router and your device can change the result by 15–20% between consecutive tests. This is entirely normal behaviour for wireless connections.
What causes large or sudden variation
Other devices using the connection simultaneously
A second device downloading a large file, running a cloud backup, or streaming 4K video will consume a significant portion of your plan's bandwidth. If you run a test while another device is saturating the connection, your result will appear much lower than your plan speed. For accurate testing, pause or disconnect other devices.
Background app activity on the test device
Automatic operating system updates, cloud sync (OneDrive, iCloud, Dropbox), antivirus scans, and browser background processes all consume bandwidth and CPU. Check your task manager or activity monitor before running a test to confirm the device is idle.
ISP congestion — time-of-day variation
If your speed is consistently good in the morning and consistently poor in the evening, your ISP's network is congested during peak hours. This is not a fault in your home network — it's a capacity problem upstream. Check ISP Rankings to see how your provider compares to others in your area.
Thermal throttling
Laptops that have been running for several hours may throttle their CPU, which affects the browser's ability to saturate a fast connection during the test. This is more common on thin laptops during the upload phase. Running the test immediately after a restart produces more consistent results.
Wi-Fi channel congestion — building density effects
In apartments or dense urban areas, dozens of Wi-Fi networks may share the same channel. Channel congestion peaks during evenings when everyone is home. A test at 2 PM may show 400 Mbps on Wi-Fi; the same test at 9 PM may show 180 Mbps purely because of shared channel interference — not your ISP. Switching to a less congested Wi-Fi channel can help significantly. How to choose the best Wi-Fi channel →
How to get consistent, comparable results
- Use Ethernet rather than Wi-Fi for benchmark testing
- Close all other browser tabs and applications before testing
- Pause cloud sync and automatic update services
- Disconnect other devices from the network, or at minimum ensure they're idle
- Run 3 tests back to back and take the median, discarding outliers
- Test at the same time of day when comparing across days
When to contact your ISP
Contact your ISP when:
- Speed on Ethernet varies by more than 25% between back-to-back tests with no background activity
- Speeds are consistently well below 50% of your plan speed at all times of day
- You see significant packet loss (above 1%) on a wired connection
- Speeds have changed significantly without any change in your setup